Traditionally, problems in the media planning space have been addressed separately from first principles on a per-problem basis. For example, a problem in the media planning space may be inserting an advertisement during transmission of a particular channel or a program where the target consumer is likely to view the advertisement. A possible approach to this problem may be a predictive solution conceived to particularly address this problem. However, one size may not fit all, and therefore it may hard to reuse the predictive solution approach for another problem in the same or another space. This is because the predictive solution approach may vary from data to data, product to product, and industry to industry. Accordingly, there is a need for a framework for generating customized predicting models, within the media planning space, as well as in other spaces where predictive solutions may be valuable.